Inaccuracies Associated with Estimating Random Measurement Errors

Abstract

This paper examines a commonly used procedure, proposed by Baecher, for separating the random measurement error associated with a particular test from the inherent spatial variability of the geological material. By comparing the two techniques of random field theory and geostatistics it is shown that the method, while well founded, requires a number of factors to be investigated before conclusions can be made regarding the random measurement error of a particular test. These factors include the nugget effect, the sample spacing, and the stationarity of the data. Two case studies are presented that are based on closely spaced data derived from a series of vertical and horizontal cone penetration tests performed in a relatively homogeneous, overconsolidated clay. The sensitivity of the results obtained from the procedure of Baecher is tested with regard to the degree of trend removed from the data and the spacing of the samples in the data set.